We investigate beam training and allocation for multiuser millimeter wave
massive MIMO systems. An orthogonal pilot based beam training scheme is first
developed to reduce the number of training times, where all users can
simultaneously perform the beam training with the base station (BS). As the
number of users increases, the same beam from the BS may point to different
users, leading to beam conflict and multiuser interference. Therefore, a
quality-of-service (QoS) constrained (QC) beam allocation scheme is proposed to
maximize the equivalent channel gain of the QoS-satisfied users, under the
premise that the number of the QoS-satisfied users without beam conflict is
maximized. To reduce the overhead of beam training, two partial beam training
schemes, an interlaced scanning (IS) and a selection probability (SP) based
schemes, are proposed. The overhead of beam training for the IS-based scheme
can be reduced by nearly half while the overhead for the SP-based scheme is
flexible. Simulation results show that the QC-based beam allocation scheme can
effectively mitigate the interference caused by the beam conflict and
significantly improve the spectral efficiency while the IS-based and SP-based
schemes significantly reduce the overhead of beam training at the cost of
sacrificing spectral efficiency a little.

本文研究了多用户毫米波巨型多输入多输出系统的波束训练和分配问题，提出了一种基于正交导频的波束训练方案和一种基于质量服务的波束分配方案，同时还针对波束训练开销的问题，提出了两种新的方案。通过模拟实验，证实了所提出的方法可以有效的增强系统性能。